Files

beabigegg d5f0df3384 feat: 新增機台稼動歷史分析功能

新增 Resource History Analysis 功能模組：
- 後端服務：query_summary、query_detail、export_csv 查詢函數
- API 路由：/api/resource/history/* 端點 (options/summary/detail/export)
- 前端頁面：KPI 卡片、OU% 趨勢圖、E10 狀態堆疊圖、熱力圖、階層式表格
- 支援時間粒度切換（日/週/月/年）與多維度篩選
- 查詢範圍擴展至 730 天（兩年）

其他改進：
- 新增 filter_cache 服務統一管理工站與型號快取
- MesApi 修復 JSON 解析失敗時誤觸重試的問題
- 新增 _safe_float() 處理 NaN 值避免 JSON 序列化錯誤
- E10 狀態分布圖表 tooltip 顯示各狀態百分比
- 新增完整測試套件（單元/整合/E2E）

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

2026-01-29 08:53:32 +08:00

2.3 KiB

Raw Blame History

MES Dashboard - Claude Code Instructions

Project Overview

MES Dashboard 是一個工廠製造執行系統的儀表板應用，使用 Flask + Oracle 資料庫 + ECharts 前端圖表。

Before Starting Any Task

Review architecture documentation: Read docs/architecture_findings.md to understand:
- Database connection management patterns
- Caching mechanisms and TTL constants
- Filter cache (workcenter/family) usage
- Frontend global components (Toast, MesApi)
- Data table filtering rules and column mappings
- E10 status definitions and OU% calculation
- Testing conventions

When Making Changes

If any of the following patterns are modified or new patterns are discovered:

Database connection or pooling approach
Caching strategy or TTL values
Global frontend components usage
Data table column names or filtering rules
New shared utilities or services
Testing conventions or setup patterns

Update docs/architecture_findings.md to reflect the changes.

Key Architecture Rules

Database

Always use mes_dashboard.core.database.read_sql_df() for queries
Never create direct connections in services
Reset db._ENGINE = None in test setUp

Caching

Use mes_dashboard.core.cache for all caching operations
Use mes_dashboard.services.filter_cache for workcenter/family lookups
Always convert WORKCENTERNAME → WORKCENTER_GROUP for display

Frontend

Toast notifications: Use Toast.warning(), Toast.error(), Toast.success() (NOT MESToast)
API calls: Use MesApi.get() with proper timeout
Array operations: Remember .reverse() modifies in-place

Data Tables

DW_MES_RESOURCE: Use PJ_ASSETSSTATUS (not ASSETSTATUS), LOCATIONNAME (not LOCATION)
DW_MES_RESOURCESTATUS_SHIFT: HISTORYID maps to RESOURCEID
DW_PJ_LOT_V: Source for WORKCENTER_GROUP mapping

SQL

Use /*+ MATERIALIZE */ hint for Oracle CTEs used multiple times
Date range: TXNDATE >= start AND TXNDATE < end + 1
Apply EQUIPMENT_TYPE_FILTER, location exclusions, asset status exclusions

Testing

Unit tests: tests/test_*_service.py
Integration tests: tests/test_*_routes.py
E2E tests: tests/e2e/test_*_e2e.py
For parallel queries (ThreadPoolExecutor), mock with function-based side_effect, not list

2.3 KiB Raw Blame History